The AI Act’s transparency rules apply as of 2 August 2026

• AI system providers must ensure that individuals are explicitly informed whenever they interact directly with an AI system (e.g., chatbots and voicebots) and that AI-generated or edited content is labelled in a machine-readable format, with some exceptions.
• In certain cases, an entity deploying the AI system must inform individuals of the use of AI for emotion recognition and biometric categorisation, label deepfake content and AI-generated or edited content published to inform the public about matters of public interest.

It is now crucial for companies to distinguish in what role they play in relation to the AI system (the provider or entity that implements the AI system), who uses the AI, what content is created, who it is intended for, and what risk of error or deception it can cause. It is the knowledge of specific ways of deploying AI in the company and understanding the relevant role that are the basic prerequisites for meeting the requirements of the AI Act.